Disaggregation is the term du jour in networking. It means taking today’s closed, proprietary, and aggregated systems and breaking them into components. In other words, disaggregating them.
Disaggregation may mean separating the hardware and software, as in the case of network functions virtualization (NFV). Or in the case of optical transport, it means separating the line system from the terminals.
In either case, the benefits are similar:
- Being able to select best-of-breed components for each part of the architecture
- Being able to innovate with one component without changing the others
- Being able to introduce competition for each component
All the above are true only if the system is both disaggregated and truly open. Hence the term DINO. DINOs seem to support the idea of disaggregation, but their actions don’t reflect their rhetoric. Let’s look at some examples.
My system is disaggregated, but all the parts must come from me
We saw this a lot in the early days of NFV. Traditional suppliers of networking gear would say the right words about transforming their router or firewall into software. But they required that you run their software on their special (and expensive) hardware instead of standard servers. A clear case of DINOs in operation!
We’ve also seen a different approach where the supplier prices inexpensive proprietary hardware to entice the customer to buy a closed solution. Once the DINO has sold the victim the proprietary hardware, it can never be used to run a competitor’s solution. Some firewall and SD-WAN suppliers like this approach.
A truly disaggregated solution comprises a set of components from different suppliers, who cooperate to deliver a solution. They know they have to win on features, performance and price – not lock-in.
You need partners with a track record of working with multiple partners to deliver an optimized solution.My system works with third-party components, but I have to approve them. Or you have to pay for them. Or their function is limited.
This is slightly better than the previous example. But the supplier has transformed their castle into a walled garden. And the gatekeeper is very strict. Examples of this include pre-approval of virtual network function software or pluggable optics before they can be used with the supposedly open base platform.
Another variation is the toll road model. You must pay each time you want to take advantage of the openness. We’ve seen open optical systems that require you to add hardware components for alien (external) wavelength support and pay hefty licensing fees for each such wavelength. Yes, it’s disaggregation – but at a cost.
A third variant of this approach is blocking or throttling of third-party components or wavelengths. For example, you might be able to use an alien wavelength in a supposedly open system. But the capabilities to monitor or protect alien wavelengths may be limited.
The net result is that innovation is severely hampered by the need to have the supplier approve each new component you want to add to the system. Not quite like loading apps on your smart phone, is it?
By using standards, a system can be both disaggregated and easily extensible. Changing a component from brand A to brand B is straightforward, because the relevant interfaces are standardized. And third-party components are not blocked.
My special component works in a disaggregated system. But that is complicated. It’s better if you get everything from me.
This is the classic bait and switch. They get you in the door with a disaggregation story, and then try to sell you the same old single-vendor solution using fear and doubt.
Those who embrace disaggregation may offer to sell a complete system. But they expect that many customers will want to buy from multiple suppliers. And they are OK with that.
My system will be disaggregated in the future. In the meantime, I’ll give you a discount.
I like to play bridge, and bridge has some great terms like slam, slough, finesse, etc. One bridge word that I really like is temporize. That is a fancy word for appearing to move forward, while actually stalling for time as you try to figure out what to do. This situation is a clear case of the supplier temporizing to try and buy some time. They aren’t ready for disaggregation now – and they may never be.
You should stick with suppliers who already have components deployed in disaggregated systems. After all, the proof is in the pudding.
It’s best to go with an ecosystem of suppliers who really believe in disaggregation
And what you really need are those with a track record of working with multiple partners to deliver an optimized solution. You must select those types of partners and insist on disaggregation. Otherwise, you may be eaten by DINOs!